1 files changed, 14 insertions(+), 0 deletions(-)

M emailbook.janet

@@ 54,6 54,7 @@ Options:
     :iso (sequence (+ "ISO" "iso") "-8859-1" (? "5"))
     :win (sequence (set "Ww") "indows-1252")
     :charset (sequence "=?" (+ :iso :win) "?Q?")}))
+(def no-ascii (peg/compile ~(range "\x80\xFF")))
 
 (defn second [xs] (get xs 1))
 (defn third [xs] (get xs 2))


@@ 228,6 229,14 @@ Options:
 
 (test (decode-iso8859-q "=?iso-8859-1?Q?M=FCller?=") "Müller")
 
+(defn invalid-ascii [str]
+  (peg/find no-ascii str))
+
+(test (invalid-ascii "hello") nil)
+(test (invalid-ascii "für") 1)
+(test (invalid-ascii "Größe") 2)
+(test (invalid-ascii "Bestätigung") 4)
+
 (defn sanitize [mailbox]
   (var sanitized mailbox)
   # ignore addresses containing noreply, no-reply or no_reply


@@ 298,6 307,11 @@ Options:
       (var results (match-mailboxes line))
       (if results
         (each mailbox results
+          # headers must use only ASCII characters (0x00 - 0x7F)
+          (when (invalid-ascii mailbox)
+            (eprintf "No valid ASCII: %s (skipped)" mailbox)
+            (break nil)) # "continue" would be useful here
+
           (def mailbox-sanitized (-> mailbox decode-utf8 decode-iso8859-q sanitize))
           (cond
             (nil? mailbox-sanitized)